HDU3255 Farming(线段树+扫描线)

本文介绍了一种通过线段树和扫描线算法解决农场播种问题的方法。该问题涉及多个矩形区域种植不同价值的种子,当矩形区域重叠时,价值较高的种子将覆盖较低的种子,最终计算所有蔬菜的总价值。

Farming

Time Limit: 12000/6000 MS (Java/Others)    Memory Limit: 65536/65536 K (Java/Others)
Total Submission(s): 1614    Accepted Submission(s): 479


Problem Description
You have a big farm, and you want to grow vegetables in it. You're too lazy to seed the seeds yourself, so you've hired n people to do the job for you.
Each person works in a rectangular piece of land, seeding one seed in one unit square. The working areas of different people may overlap, so one unit square can be seeded several times. However, due to limited space, different seeds in one square fight each other -- finally, the most powerful seed wins. If there are several "most powerful" seeds, one of them win (it does not matter which one wins).

There are m kinds of seeds. Different seeds grow up into different vegetables and sells for different prices.
As a rule, more powerful seeds always grow up into more expensive vegetables.
Your task is to calculate how much money will you get, by selling all the vegetables in the whole farm.
 

Input
The first line contains a single integer T (T <= 10), the number of test cases.
Each case begins with two integers n, m (1 <= n <= 30000, 1 <= m <= 3).
The next line contains m distinct positive integers p i (1 <= p i <= 100), the prices of each kind of vegetable.
The vegetables (and their corresponding seeds) are numbered 1 to m in the order they appear in the input.
Each of the following n lines contains five integers x1, y1, x2, y2, s, indicating a working seeded a rectangular area with lower-left corner (x1,y1), upper-right corner (x2,y2), with the s-th kind of seed.
All of x1, y1, x2, y2 will be no larger than 10 6 in their absolute values.
 

Output
For each test case, print the case number and your final income.
 

Sample Input
  
2 1 1 25 0 0 10 10 1 2 2 5 2 0 0 2 1 1 1 0 3 2 2
 

Sample Output
  
Case 1: 2500 Case 2: 16
题意:给出n个矩形,每块矩形种不同价值的种子,矩形会出现相交情况,这时候价值高的种子会把价值低的种子淘汰,问最后得到全部价值。
思路:
如果线段树+扫描线可以求覆盖面积,那么这题就是求体积并,试想把种子的价值当作高,那么按高度升序排序后,分段求体积,选出高度不小于h[i]的矩形插入线段树求当前底面积,每次就相当于求一边覆盖面积(其实就是当前底面积,那么久相当同一块矩形里,最高的高(价值最大的种子)会被当作高去求体积),求得的覆盖面积*(h[i+1]-h[i])。那么把每次求得的分段体积加起来就是总的体积了。
线段树+扫描线求覆盖面积: 点击打开链接

#include <stdio.h>
#include <string.h>
#include <algorithm>
using namespace std;
#define maxn 35000
#define ll long long int
ll sum[maxn<<2];
int mark[maxn<<2], val[5], Hash[maxn<<1];
struct seg{
    int l, r, h, c, val;
    seg(){}
    seg(int x1, int x2, int y, int v, int d):l(x1),r(x2),h(y),val(v),c(d){}
    bool operator <(const seg &a)const{
        return h<a.h;
    }
}s[maxn<<1], tmp[maxn<<1];
int Search(int key, int a[], int right){
    int l = 0, r = right-1;
    while(l <= r){
        int mid = (l+r)>>1;
        if(a[mid] == key) return mid;
        if(a[mid] > key) r = mid-1;
        else l = mid+1;
    }
    return -1;
}
void upfather(int n, int left, int right){
    if(mark[n]) sum[n] = Hash[right+1] - Hash[left];
    else if(left == right) sum[n] = 0;
    else sum[n] = sum[n<<1] + sum[n<<1|1];
}
void update(int L, int R, int rt, int d, int left, int right){
    if(L<=left&&right<=R){
        mark[rt]+=d;
        upfather(rt, left, right);
        return;
    }
    int mid = (left+right)>>1;
    if(L<=mid) update(L, R, rt<<1, d, left, mid);
    if(mid<R) update(L, R, rt<<1|1, d, mid+1, right);
    upfather(rt, left, right);
}
int main()
{
    int T, k, i, j, n, m, x1, x2, y1, y2, Case, v, p;
    scanf("%d", &T);
    Case = 1;
    while(T--){
        scanf("%d %d", &n, &m);
        for(i = 1;i <= m;i++) scanf("%d", &val[i]);
        k = 0;
        for(i = 0;i < n;i++){
            scanf("%d %d %d %d %d", &x1, &y1, &x2, &y2, &v);
            Hash[k]= x1;
            s[k++] = seg(x1, x2, y1, val[v], 1);
            Hash[k]= x2;
            s[k++] = seg(x1, x2, y2, val[v], -1);
        }
        val[0] = 0;
        sort(Hash, Hash+k);
        sort(s, s+k);
        sort(val, val+m+1);
        p = 1;
        for(i = 1;i < k;i++)
            if(Hash[i] != Hash[i-1])
                Hash[p++] = Hash[i];
        ll ans = 0;
        for(i = 1;i <= m;i++){
            int c = 0;
            for(j = 0;j < k;j++)
                if(s[j].val > val[i-1])
                    tmp[c++] = s[j];

            for(j = 0;j < c;j++){
                int L = Search(tmp[j].l, Hash, p);
                int R = Search(tmp[j].r, Hash, p)-1;
                //printf("%d %d\n", tmp[j].l, tmp[j].r);
                update(L, R, 1, tmp[j].c, 0, p-1);
                //printf("%I64d\n", sum[1]);
                ans += sum[1]*(val[i] - val[i-1])*(tmp[j+1].h - tmp[j].h);
                //printf("%I64d\n", sum[1]);
            }
        }
        printf("Case %d: %I64d\n", Case++, ans);
    }
}


<think> 嗯...用户这次问的是扫描线算法中线段树是否动态开点。看来用户正在学习计算几何的高级数据结构应用,可能是个OI选手或者算法爱好者。 从用户引用的两篇文献来看,第一篇详细描述了扫描线配合线段树求矩形面积并的标准流程,第二篇则涉及矩形覆盖最大点数的变种问题。用户能提出"动态开点"这样的专业术语,说明已经对线段树的常见优化方式有所了解。 关于动态开点问题,需要分场景讨论:在标准的离散化扫描线中,由于横坐标已离散化为固定区间,线段树可以预先建树无需动态开点;但在处理浮点数或值域过大的情况时,动态开点就成为必要选择。这个区别很关键,用户可能正在处理特殊场景才会这么问。 考虑到用户之前关注过线段树维护区间覆盖的技术细节,后续可以补充离散化与动态开点的性能对比。不过当前问题需要先明确回答核心疑问——常规实现不用动态开点,特殊情况才需要。 </think> 在扫描线算法中,线段树**通常不需要动态开点**,但具体实现取决于问题需求和坐标系特征: ### 📍 标准实现(非动态开点) 1. **离散化处理** 扫描线算法的关键步骤是将所有坐标点在 $x$$-轴上离散化(如图1所示): $$ \text{步骤:收集坐标} \rightarrow \text{排序} \rightarrow \text{去重} \rightarrow \text{建立映射} $$ 离散化后,线段树管理的区间数量被压缩为 $O(n)$ 级别($n$ 为坐标点数),因此可直接**预建满二叉树**,无需动态开点[^1]。 2. **空间复杂度优化** 预建树的空间复杂度为 $O(n)$,例如: - 原始坐标范围:$[1, 10^9]$ - 离散化后区间数:$2000$(若 $n=1000$) 此时只需开 $4 \times 2000 = 8000$ 的节点数组,完全可行[^1]。 3. **维护信息** - 节点存储 **区间覆盖次数** 和 **有效长度** - 更新操作:区间加减(矩形入边 `+1`,出边 `-1`) - 查询操作:直接取根节点有效长度 ### ⚡ 需动态开点的情况 | **场景** | **原因** | **示例** | |-----------------------|--------------------------------------------------------------------------|---------------------------| | **值域极大且无法离散化** | 浮点数坐标或值域超限(如 $[1, 10^{18}]$) | 地理信息系统(GIS)中的浮点计算 | | **强制在线查询** | 矩形边界实时插入,无法预离散化 | 流式数据处理场景 | | **内存严格受限** | $n$ 极大(如 $>10^6$),离散化后区间仍过多,需节省空间 | 嵌入式设备上的算法部署 | ### 🌰 实例对比 1. **标准问题(矩形面积并)** ```python # 离散化坐标 xs = sorted(set(x_coords)) tree = SegmentTree([0]*4*len(xs)) # 预建树 ``` 2. **动态开点问题(浮点数覆盖)** ```python class Node: def __init__(self): self.cover = 0 self.len = 0.0 self.lson = None # 动态创建子节点 self.rson = None ``` ### ✅ 结论 - **大多数场景**(如引用[1][2]的矩形问题):离散化 + 静态线段树 ✅ - **特殊场景**(浮点数/在线查询):动态开点线段树 ✅ 若处理整数坐标系且可离散化,推荐静态实现,效率更高且编码简单[^1][^2]。 --- ### ❓相关问题 1. 扫描线算法中如何处理浮点数坐标的精度问题? 2. 离散化过程如何避免哈希冲突导致的查询错误? 3. 动态开点线段树扫描线中的更新操作时间复杂度如何保证? 4. 矩形覆盖最大值问题(如HDU 5091)如何转化为扫描线模型?[^2] [^1]: 线段树应用——扫描线,离散化坐标与静态建树 [^2]: HDU 5091 矩形覆盖问题中的扫描线转化技巧
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值